Preserve br tag in ttml

ttml.go

@@ -197,6 +197,39 @@ func (i *TTMLInItems) UnmarshalXML(d *xml.Decoder, start xml.StartElement) (err
 	return nil
 }
 
+// handleBrTokenReader is used only for decoding TTMLInItems, do not use it anywhere else
+type handleBrTokenReader struct {
+	xml.Decoder
+	holdingToken xml.Token
+}
+
+// Token implements the TokenReader interface, when it meets the "br" tag, it will hold the token and return a newline
+// instead. This is to work around the fact that the go xml unmarshaler will ignore the "br" tag if it's within a
+// character data field.
+func (r *handleBrTokenReader) Token() (xml.Token, error) {
+	if r.holdingToken != nil {
+		returnToken := r.holdingToken
+		r.holdingToken = nil
+		return returnToken, nil
+	}
+
+	t, err := r.Decoder.Token()
+	if err != nil {
+		return nil, err
+	}
+
+	if se, ok := t.(xml.StartElement); ok && strings.ToLower(se.Name.Local) == "br" {
+		r.holdingToken = t
+		return xml.CharData("\n"), nil
+	}
+
+	return t, nil
+}
+
+func newHandleBrTokenReader(r io.Reader) xml.TokenReader {
+	return &handleBrTokenReader{Decoder: *xml.NewDecoder(r), holdingToken: nil}
+}
+
 // TTMLInItem represents an input TTML item
 type TTMLInItem struct {
 	Style string `xml:"style,attr,omitempty"`
@@ -380,7 +413,8 @@ func ReadFromTTML(i io.Reader) (o *Subtitles, err error) {
 
 		// Unmarshal items
 		var items = TTMLInItems{}
-		if err = xml.Unmarshal([]byte("<span>"+ts.Items+"</span>"), &items); err != nil {
+		decoder := xml.NewTokenDecoder(newHandleBrTokenReader(strings.NewReader("<p>" + ts.Items + "</p>")))
+		if err = decoder.Decode(&items); err != nil {
 			err = fmt.Errorf("astisub: unmarshaling items failed: %w", err)
 			return
 		}

ttml_test.go

@@ -2,10 +2,12 @@ package astisub_test
 
 import (
 	"bytes"
-	"github.com/asticode/go-astikit"
 	"io/ioutil"
+	"strings"
 	"testing"
 
+	"github.com/asticode/go-astikit"
+
 	"github.com/asticode/go-astisub"
 	"github.com/stretchr/testify/assert"
 )
@@ -50,3 +52,19 @@ func TestTTML(t *testing.T) {
 	assert.NoError(t, err)
 	assert.Equal(t, string(c), w.String())
 }
+
+func TestBreakLineHandling(t *testing.T) {
+	// Open
+	s, err := astisub.OpenFile("./testdata/example-with-breaklines-in.ttml")
+	assert.NoError(t, err)
+
+	// Write
+	w := &bytes.Buffer{}
+	err = s.WriteToTTML(w)
+	assert.NoError(t, err)
+
+	c, err := ioutil.ReadFile("./testdata/example-with-breaklines-out.ttml")
+	assert.NoError(t, err)
+
+	assert.Equal(t, strings.TrimSpace(string(c)), strings.TrimSpace(w.String()))
+}